Genetic Algorithm Optimized Probabilistic Maintenance Scheduling

The Air Force does not have an optimal means for performing maintenance on jet aircraft engines. Although engine repair can be based on scheduled removal, most removals occur after engine failure. The performance criteria for optimal maintenance involve, at least, maximizing the percentage of mission capable aircraft over a given period of time while minimizing the cost per engine flight hour, specific maintenance, and logistics. The space of possible repair strategies can be represented on a computer and searched effectively using an evolutionary algorithm. Natural Selection Inc. and subcontractor GE GRC will adapt an existing available simulation and design an evolutionary approach that considers the expected life of the engine parts, the random failure rates, operational readiness requirements, required scheduled maintenance constraints, the costs of maintenance actions, and the cost of replacement including lead time considerations. Phase I experiments will provide a basis for assessing the transition requirements for Phase II to construct a system to execute in a prototype operation to minimize the cost per engine flight hour based on available cost data. The process will be transitioned to service both the Air Force aircraft, as well as those of other services, and commercial customers for GE.